De-identification Methods for Soft and Non-biometric Identifiers is a research project at the Department of Electronics, Microelectronics, Computer and Intelligent Systems, Faculty of Electrical Engineering and Computing, University of Zagreb, Croatia. The project is funded by the Croatian Science Foundation (UIP-11-2013-1544 DeMSI)

About the project

Advances in audio and video recording devices, web technology and signal processing have highly facilitated the efficacy of audio and video acquisition. This capability is now widely exploited in a variety of scenarios to obtain audio-video recordings of people, raising serious privacy concerns. The protection of sensitive private data is required by EU’s Data Protection Directive (95/46/EC).

De-identification in multimedia content can be defined as the process of concealing identities of individuals captured in a given set of data (images, video, audio, text), for the purpose of protecting their privacy. Soft biometric and non-biometric characteristics such as tattoos, birthmarks scars or dressing style also carry potentially identity-revealing information and have to be dealt with in the process of de-identification. This is an interdisciplinary challenge, involving such scientific areas as image processing, video tracking and biometrics.

This project aims to development of novel and robust methods for de-identification of soft and non-biometric identifiers in videos and images.

Project objectives

• To determine the classes of soft biometric and non-biometric identifiers that are normally present in image and video contents.
• To specify approaches to characterizing the correspondence between the choice of de-identification methods and the given scenarios and environments.
• To provide innovative solutions for concealing, or removal of, soft biometric and non-biometric identifiers while preserving data utility and/or naturalness.
• To investigate reversible de-identification and to provide a thorough analysis of security risks of reversible de-identification.